Purpose of this notebook

Query the results database for baseline results that are best displayed in a table (as opposed to a graph)



In [2]:

    
%cd ~/NetBeansProjects/ExpLosion/
from notebooks.common_imports import *
import numpy as np









    



/Users/miroslavbatchkarov/NetBeansProjects/ExpLosion



In [23]:

    
# bag-of-NPs results
for e in Experiment.objects.filter(expansions__decode_handler='BaseFeatureHandler'):
    mean, low, high, _ = get_ci(e.id)
    print(e.id, e.document_features_tr, e.document_features_ev, e.labelled, mean, high, low, (high-low)/2)









    



5 J+N+AN+NN AN+NN amazon_grouped-tagged 0.845399832355 0.853062092978 0.837671996905 0.00769504803662
6 J+N+AN+NN AN+NN reuters21578/r8-tagged-grouped 0.927997311828 0.944559811828 0.91095094086 0.0168044354839
7 J+N+AN+NN J+N+AN+NN amazon_grouped-tagged 0.897175814536 0.903261904762 0.891037593985 0.00611215538847
247 J+N+V+SVO SVO amazon_grouped-tagged 0.731692939245 0.752873563218 0.706689244663 0.0230921592775



In [22]:

    
# random vectors/neighbours
means = []
for r in Experiment.objects.filter(expansions__vectors__algorithm__startswith='random_'):
    mean, low, high, _ = get_ci(r.id)
    print(r.id, r.labelled, r.expansions.vectors.algorithm, mean, high, low, (high-low)/2)









    



1 amazon_grouped-tagged random_neigh 0.21804459735 0.226905262487 0.209670616718 0.00861732288481
2 amazon_grouped-tagged random_vect 0.218249108053 0.227223814985 0.209600535168 0.00881163990826
3 reuters21578/r8-tagged-grouped random_neigh 0.502913612565 0.533383507853 0.476763743455 0.028309882199
4 reuters21578/r8-tagged-grouped random_vect 0.502280104712 0.537964659686 0.465307591623 0.0363285340314



In [ ]: